摘要：The binary classification of coding and non-coding genes is simplified near to 50 years. Genome-wide transcriptome studies have revealed that there exist tens of thousands of long non-coding RNAs (lncRNAs), while the functions are being uncovered slowly. Accurate identification of lncRNAs is the initial step to the systematic characterization of lncRNAs. The diversity of transcription patterns for lncRNAs challenges the available non-coding RNA prediction algorithms. Until now, prediction of lncRNAs mostly relies on genomic sequence and cross-species alignment information. Here, we introduce the main strategies that can discriminate lncRNA from protein-coding transcripts. Especially, recently available machine learning algorithms are shown efficient to the rapid and accurate identification of lncRNAs from a large number of putative lncRNAs based on transcriptome assembled transcripts, which would provide the basis of understanding of lncRNA biology.
